Article Text
Abstract
Background Fatty liver disease (FLD), of either alcoholic (AFLD) or non-alcoholic (NAFLD) aetiologies, is characterised histologically by features that are lobulocentric that typically affect zone 3, in contrast with portal-based inflammation characteristics of other forms of chronic liver disease.
Aim The authors aimed to determine the prevalence and significance of portal inflammation in a cohort of adults diagnosed as having AFLD and NAFLD.
Methods From a histology database, the authors identified 160 patients with biopsy proven AFLD and 214 with NAFLD diagnosed between 1991 and 2001. All liver biopsies were reviewed by one pathologist to evaluate various histological features.
Results More than a mild degree of portal inflammation was found in 47% of AFLD compared with 30% in NAFLD subgroup. A higher degree of portal inflammation in AFLD was associated with a higher mean corpuscular volume, gamma glutamyl transferase (GGT) and alkaline phosphatase, and a lower platelet count and albumin, whereas in NAFLD this was associated with age, presence of diabetes, hypertension, body mass index (BMI), higher fasting blood glucose, cholesterol, alanine transaminase, GGT and ferritin. On regression analysis, portal inflammation was associated with the severity of steatosis (p=0.005), presence of ballooning (p=0.030) in NAFLD and severity of fibrosis in both AFLD and NAFLD (p<0.001).
Conclusions Portal inflammation is a common component of histological spectrum of both AFLD and NAFLD. In both conditions portal inflammation is associated with clinical and histological features suggestive of advanced disease.
- Fatty liver disease
- alcoholic
- non-alcoholic
- portal inflammation
- inflammation
- liver
Statistics from Altmetric.com
Introduction
Fatty liver disease (FLD) can be due to alcohol excess (AFLD) or non-alcoholic aetiology (NAFLD). The term NAFLD is used to describe the liver findings in patients with FLD in the absence of excess alcohol consumption.1 NAFLD is an important complication of the metabolic syndrome, including central obesity, insulin resistance, hypertension and hyperlipidaemia. NAFLD is now a common cause of chronic liver disease.2–4 As can be expected from their description, the histologies of AFLD and NAFLD are similar, and the distinction between the two is ultimately clinically derived.5 6 However, natural histories suggest that AFLD may progress more frequently than NAFLD.7 8
Histologically, FLD is a spectrum, with the term steatohepatitis (SH) being used to define changes which indicate more severe and potentially progressive liver disease. Although the best definition of steatohepatitis remains unsettled, it can be defined as liver disease characterised by steatosis with evidence of liver cell injury (ballooning, necrosis and apoptosis) and inflammation with or without fibrosis mainly affecting zone 3.9–11 It is recognised that the histological changes in AFLD or NAFLD are lobulocentric (typically affecting zone 3), and they are sufficiently distinct from the portal-based lesions of chronic hepatitis such as viral and autoimmune hepatitis. Portal-based chronic inflammation in FLD, although relatively common, still remains an underappreciated lesion with debatable significance.12–14 The degree of portal inflammation was considered as one of the criteria of grading of non-alcoholic steatohepatitis (NASH) by the Brunt scoring system in 1999,11 13 but it was not included in the NASH Clinical Research Network (CRN) scoring system in 2005.15 Some authors have reported that portal inflammation in NAFLD, particularly if disproportionately high, may be an indication of the possibility of another, concurrent chronic ‘portal-based’ liver disease.16–18 In a recent study by the CRN group, Brunt et al14 concluded that more than mild portal inflammation in liver biopsies of untreated NAFLD may be considered a marker of advanced disease. Moreover, data on the significance of portal inflammation in AFLD are lacking.19 20 Therefore, the aims of this study are to determine the prevalence of portal inflammation as well as its clinical and histological correlates in a large series of adults with liver biopsy evidence of fatty liver disease of both alcoholic (AFLD) and non-alcoholic (NAFLD) aetiologies, in a single institution.
Materials and methods
Subjects and clinical data collection
The pathology database at Queen's Medical Centre, Nottingham University was searched for cases coded as ‘fatty liver’ from January 1991 to December 2001. Using clinical information and laboratory tests, we excluded all those cases where drug therapy could have contributed to accumulation of fat (such as methotrexate, steroids, amiodarone, tetracycline, etc) as well as those associated with other specific liver diseases such as hepatitis B and C viral infection, haemochromatosis, auto-immune liver diseases, Wilson's disease, and alpha-1 antitrypsin deficiency. We identified adult subjects who were more than 18 years of age with biopsy-proven FLD. FLD was initially defined by the minimal criteria of hepatic steatosis (>5%) with or without lobular inflammation, Mallory–Denk bodies, cytological ballooning and fibrosis.1 9 21 22
The following clinical data were collected:
Patients characteristics including patient's age, gender, body weight and height (body mass index), presenting symptoms, past medication history, presence of diabetes, hypertension (including systolic and diastolic blood pressure), and the level of alcohol consumption (units per week) and its duration were also recorded.
Results of liver chemistry (serum alanine transferase (ALT), alkaline phosphatase (ALP) and gamma-glutamyltransferase (GGT) in addition to serum bilirubin and albumin).
Other blood tests including platelet counts, mean corpuscular volume (MCV), lipid profile (cholesterol level, low (LDL)/high-density lipoproteins (HDL) and triglycerides), C-reactive protein, ferritin level, creatinine, fasting blood glucose and glycated haemoglobin (haemoglobin A1c).
Immunological markers including immunoglobulins IgA, IgG and IgM levels, and liver-disease-related autoantibodies (such as anti-nuclear, smooth muscle, liver kidney microsomal and antimitochondrial antibodies).
Histopathology
In our institution, sections of liver biopsies are routinely stained with haematoxylin–eosin, periodic acid Schiff, periodic acid Schiff-digested, reticulin, Picro-sirius red and/or Masson trichrome reagents, elastic tissue stain and iron stain. In the current study, all biopsies were reviewed by a hepatopathologist (PVK) without any knowledge of the clinical information. The length and width of the biopsies were recorded. The presence and severity of histological features were determined in a standardised way and scored semiquantitively from 0 to 4 as previously described by Promrat et al,23 and have been used in clinical trial.24 The scoring system was used for the entire spectrum of FLD without prior knowledge as to whether it was AFLD or NAFLD. These included ballooning degeneration of hepatocytes, parenchymal inflammation, portal inflammation, fibrosis and Mallory–Denk bodies.11 Portal inflammation was graded as 0 (none), 1 (mild, sprinkling of lymphocytes in some or all portal tracts), 2 (mild to moderate, denser lymphocytic infiltrate in some or most portal tracts), 3 (moderate to severe, dense lymphocytic infiltrate in most or all portal tracts) or 4 (severe, dense lymphocytic infiltrate in almost all or all portal tracts). Steatosis was graded as 1 (5–25%), 2 (26–50%), 3 (51–75%) and 4 (76–100%). Fibrosis was scored as 0 (no fibrosis), 1 (sinusoidal/pericellular), 2 (periportal), 3 (bridging) and 4 (diffuse).23 In addition, perisinusoidal/pericellular and portal fibrosis were assessed and scored separately as 0–4.25 The presence of oxyphilic change in hepatocytes, which is a form of cellular injury/degeneration due to accumulation of mitochondria frequently seen in severe alcoholic liver disease,26 was scored as absent or present. The degree and extent of fibrosis were evaluated on the trichrome stained slides according to the Kleiner system.15
The degree of alcohol consumption as estimated at the time of presentation to the hospital was extracted from the case notes. If the subject consumed excess alcohol (units/week >14 for females, >21 for males) then it is considered alcoholic fatty liver disease, whereas those who consumed less than that amount of alcohol were classified as non-alcoholic fatty liver disease.11 27 In this study, FLD which scored positively for cellular injury (ballooning/oncocytic/oxyphilic change/necrosis) and/or fibrosis was regarded as steatohepatitis.
Data analysis
Statistical analysis was performed using SPSS 15.0 statistical software (SPSS, Chicago, Illinois). The means, standard deviations, medians, ranges and percentages were reported for various demographic and clinical features. In characterising the relationships between portal inflammation and clinical and histopathological features, the Spearman rank correlation, χ2 test and Mann–Whitney U test were used. To identify relationships between the severity of portal inflammation and laboratory/histological features, the Student t test, linear regression and the Welch test for unequal variances were used. The associations between portal inflammation and other clinicopathological variables were analysed using the degree of portal inflammation scored as five categories (0–4) as well as when dichotomised (as absent (0) or mild (1) versus more than mild (2–4)).14 20 A p value <0.05 was considered significant. However, with multiple statistical comparisons, a p value <0.01 was considered significant.
Results
Demographic and clinical features
Out of 418 liver biopsies coded as fatty liver, 374 cases that met the criteria for inclusion in this study were included (adults with fatty liver without other concomitant chronic liver disease). Fourteen cases were further excluded due to absent or insufficient biopsy material, and therefore, 360 cases (263 males, 97 females) were included in the final analysis. These included 154 (43%) patients with AFLD and 206 (57%) with NAFLD. The median age of patients was 49 years (range: 20–80). The most frequent presentation was abnormal liver function tests with or without other symptoms. Pain was the presenting symptom in 33 (9%) patients. The median length of biopsies was 19.5 mm (range 5–70), and the median width was 1 mm (range 0.4–1.5); 296 (82%) were regarded as steatohepatitis (SH), while 64 cases did not show any evidence of parenchymal injury or fibrosis
Characteristics of AFLD and NAFLD
The clinical and histological characteristics of the whole cohort and the difference between AFLD and NAFLD subgroups are detailed in tables 1 and 2. AFLD patients were more likely to be male but less likely to be diabetic than patients with NAFLD. Patients with AFLD more often presented with abnormal liver function tests (elevated GGT, ALP and total bilirubin and reduced albumin), lower platelet count, higher ferritin and IgA levels when compared with NAFLD patients.
All histological features of FLD apart from steatosis itself were significantly more severe in the AFLD group, and the stage was also more advanced in this group. More than a mild degree of portal inflammation was found in 47% of AFLD compared with 30% in NAFLD subgroup (p=0.001). Regression analysis showed that fibrosis stage is the most significant histological feature associated with AFLD subgroup (p<0.001) followed by oxyphilic changes (p=0.028).
Correlation between portal inflammation and other clinicopathological variables
NAFLD subgroup
Patients with more than mild portal inflammation were older, more likely to be diabetic and more likely to have high ALT and ferritin levels and LDL levels. There was no association between portal inflammation and levels of albumin, alkaline phosphatase and immunoglobulins, or autoantibody positivity (table 3).
An increasing degree of portal inflammation was associated with other features which indicate the severity of the disease including advanced grade of steatosis, lobular inflammation, liver cell injury (ballooning degeneration) and the presence and extent of fibrosis including pericellular and portal fibrosis (p<0.001) and the presence of SH (table 4). Regression analysis showed an association between portal inflammation and severity of fibrosis (p<0.001), degree of steatosis (p=0.005) and the presence of ballooning degeneration (p=0.030) while lobular inflammation, Mallory hyaline and iron deposition were not significant (p>0.05).
AFLD subgroup
More than mild portal inflammation was more frequently associated with high alkaline phosphatase and ferritin, and low albumin and high-density lipoprotein levels. Similar to the NAFLD subgroup, there were associations between increasing degree of portal inflammation and other features which indicate the severity of the disease including advanced stage of lobular inflammation, ballooning degeneration, steatosis, oxyphilic changes, fibrosis, including portal fibrosis (p<0.001) and the presence of SH (table 4). Regression analysis showed an association between portal inflammation and severity of fibrosis (p<0.001) while other histological features were not significant (p>0.05).
Similar associations were found when portal inflammation was analysed as categories (0–4) in both NAFLD and AFLD. There was a correlation between the length of the biopsy and portal inflammation in NAFLD (p<0.001) but not in AFLD (p=0.21). Similar correlations were also found between the length of the biopsy and both lobular inflammation and fibrosis but not with other variables. No association was found between immunoglobulins (IgA, IgG or IgM) or autoantibody positivity and other histological findings apart from association between IgA and degree of steatosis and stage of fibrosis in NAFLD (p<0.001). Autoantibody positivity was associated with female sex (χ2=7.3, p=0.007) in the whole series. However, when analysed separately, this association with female sex was found in NAFLD (χ2=9.2, p=0.002) but not in the AFLD subgroup (χ2=0.01, p=0.983).
Discussion
The histological spectrum of FLD includes steatosis, hepatocellular injury (ballooning degeneration, apoptosis or Mallory bodies), inflammation and fibrosis of varying degree. As AFLD and NAFLD share a similar histological spectrum, the distinction between the two conditions remains clinically based, almost entirely on a history of alcohol intake. Key characteristic histological features in AFLD and NAFLD have been noted in centrilobular location which may be consistent with the metabolic zonation within the liver acinus,28 and periportal changes have received relatively less attention. Our study shows that portal inflammation is seen in the majority of subjects with AFLD (71%) as well as NAFLD (63%). Our findings are consistent with two recent studies; one found portal inflammation in 83% of 728 adults and 90% of 205 children with NAFLD,14 and another found portal inflammation in 40% of 200 adults AFLD.20
In NAFLD, portal inflammation has previously been noted in certain clinical settings such as in children,29 those with concurrent chronic liver disease30 or following treatment of NAFLD especially with glitazones.31 Portal-based pathology has also been described as distinct pathological entities in paediatric subjects such as type 2 NASH without ballooning and perisinusoidal changes29 and isolated portal fibrosis seen in morbidly obese subjects.32 Our cohort included well-characterised patients with FLD from which other causes of chronic liver disease and drug-induced fatty liver disease had been excluded; hence, our findings confirm with another recent retrospective series33 that portal inflammation is a common component of histological spectrum of both AFLD and NAFLD.
In patients with NAFLD, more than mild portal inflammation correlated with several clinical parameters such as age, diabetes, hypertension, BMI as well as laboratory parameters such as fasting blood glucose, ALT, total cholesterol, LDL and ferritin levels. Previous reports have described the association between advanced histological changes such as fibrosis in NAFLD and many of these clinical parameters such as age, BMI,34–36 presence of diabetes,34 35 37 hypertension,37 ALT and triglyceride levels36 in different cohorts of NAFLD patients. Our findings are also in agreement with a recent report from the USA,14 which found more than mild portal inflammation to be associated with age, female gender, BMI, diabetes and hypertension, although, unlike our study, no association between portal inflammation and ALT levels was found. In contrast to NAFLD, none of the factors such as age, presence of diabetes, obesity and hypertension was associated with more than mild portal inflammation in AFLD cases. Instead, in AFLD, lower albumin levels, lower platelet counts, higher MCV and ferritin levels were associated with significant portal inflammation. All these factors are surrogate markers of advanced AFLD; hence similar to NAFLD more than mild portal inflammation in AFLD implies advanced histological changes. However, although we found ferritin levels to be associated with portal inflammation in both AFLD and NAFLD, there are conflicting reports regarding association of hepatic fibrosis with serum ferritin levels in NAFLD with data both confirming38 and denying the association.39
In this study, portal inflammation was associated with features of advanced histological changes including the extent of steatosis, presence and extent of lobular inflammation, liver cell injury, fibrosis and the presence of SH. However, we cannot confirm whether portal inflammation is simply a marker of severity in AFLD/NAFLD or contributes to pathogenesis of this progressive disease. Previous studies have suggested that small intestinal bacterial overgrowth, intestinal permeability and gut-derived endotoxaemia may play a role in the pathogenesis of AFLD40 and NAFLD.41 42 Consistent with such a hypothesis, we found that raised IgA levels were associated with the higher degree of steatosis and stage of fibrosis (p<0.001). Alternatively, it can be hypothesised that activation of the innate immune system and increased release of pro-inflammatory cytokines from portal tract inflammatory cells may lead to perpetuation of inflammation, hepatocyte damage and hence advanced disease in both AFLD and NAFLD.43–45 In our study, autoantibody positivity was strongly associated with female gender in the NAFLD cohort, thus supporting the view that activation of immune system may contribute in pathogenesis. However, there was no association between portal inflammation and autoantibodies or immunoglobulins in AFLD or NAFLD as a group. A previous study investigating the determinants of progressive fibrosis in NAFLD showed that portal inflammation was independently and strongly (odds ratio of 3) associated with fibrosis.12 These authors hypothesised that as the primary pathway of replacement of necrotic and apoptotic hepatocytes through replication of adjacent hepatocytes is impaired in NAFLD, a secondary proliferative pathway, the so-called ‘ductular reaction’, is activated, hence explaining the association of portal inflammation with progressive fibrosis.12 A similar association between portal inflammation and fibrosis has also been reported in AFLD.20 In a recent study, Argo et al46 identified portal inflammation as a predictor of fibrosis progression as well as a marker of disease severity in NAFLD. However, only long-term natural history studies including a large cohort of subjects with repeat biopsies at follow-up will be able to confirm the significance of portal inflammation in NAFLD.
Overall, in our study, AFLD patients as a group had more advanced histological features compared with those with NAFLD at the time of biopsy. Both frequency and degree of inflammation (lobular and portal), evidence of hepatocellur injury (ballooning and Mallory–Denk bodies) and fibrosis (presence/extent of pericellular fibrosis and the overall fibrosis stage) were significantly higher in the AFLD group. Consistent with these, AFLD group had clinical evidence of more advanced disease as shown by higher bilirubin, lower albumin and lower platelet counts, despite the fact that there was no difference in the mean age between the two groups. Although AFLD and NAFLD both have a similar histological spectrum and hence are inseparable based on histopathology alone, the two conditions have a different natural history. This was demonstrated by two landmark studies of natural history of fatty liver.7 8 In these studies, two retrospective cohorts of pure fatty liver were followed up for a period of 10 years. While 2.5% (1/40) of NAFLD developed mild fibrosis on follow-up biopsy,7 18% of those with AFLD developed fibrosis; including 10% with cirrhosis.8 This points to a more rapid progression of AFLD when compared with NAFLD under the circumstances where there is persistent exposure to risk factors. Therefore, it is unsurprising that the cohort with AFLD have more advanced liver disease compared with the NAFLD cohort. We acknowledge that our study design has some inherent limitations. Patients forming these cohorts were identified from the liver biopsy database, and biopsies were performed based on clinical indications rather than any predetermined criteria. Therefore, the threshold for biopsy may have varied depending on clinical circumstances and the clinicians requesting liver biopsy. It is often difficult to draw a line between AFLD and NAFLD. Alcoholic history may be difficult to obtain and quantify. Alcohol consumption may coexist with factors of metabolic syndrome.
We conclude that portal inflammation is a common component of the histological spectrum of both AFLD and NAFLD. In both conditions, portal inflammation is associated with clinical and histological features suggestive of advanced disease.
Take-home messages
Alcoholic (AFLD) and non-alcoholic (NAFLD) fatty liver disease share a similar histological spectrum, and the distinction between the two conditions remains clinically based.
Histological features of advanced liver disease such as hepatocellular injury, inflammation and fibrosis were significantly more severe in AFLD than in NAFLD.
Although the key characteristic histological features in AFLD and NAFLD have been noted in centrilobular location, this study demonstrates that portal inflammation is a common component of the histological spectrum of both AFLD and NAFLD.
More than mild degree of portal inflammation is found in approximately half of AFLD compared with a third in NAFLD subgroup.
In both conditions, portal inflammation is associated with clinical and histological features suggestive of advanced disease including severity of fibrosis.
No significant association is found between level of immunoglobulins or autoantibody positivity and histological findings in FLD.
References
Footnotes
PVK and GPA contributed equally.
Competing interests None.
Ethics approval Ethics approval was provided by the Nottingham Ethics Committee.
Provenance and peer review Not commissioned; externally peer reviewed.